GoodSoundClub - Romy the Cat's Audio Site

In the Forum: Audio Discussions
In the Thread: It’s mad, mad, mad... electricity.

Post Subject: Funny electricityPosted by JOHNR on: 1/27/2016

Paul,
It would be improper for me to write down what I have worked on over almost 40 years in the telcoms business.
WRT a CV for example, I have in the past written it all down and the reviewers do not believe it. Eventually, i stopped bothering about it.
.
To give a measure, when I was invited to join LinkedIn, I had 8000 users who wanted to communicate. Needless to say, I rapidly abandoned that.

My focus over the past 10 years has been to determine where noise comes from on a triode based amplifier. This is where I developed the practice of diverting such noise sources and to feed them into a radio frequency ground path (RFG) to a dedicated ground. My 48 volt power source operates with 2 ground paths namely, safety and RFG. I picked up on this methodology whilst tracking down a current loop on a switching system that had 80 amperes of noise in the grounding system. The manufacturer disbelieved what I showed them and I had the grounding system changed about whilst in service. The current fell to 0.3 amps and they were stunned to say the least.

Decided to construct a 300B amp and learned from that.

Transformers, exotic does not mean much to me. I use amorphous cores for the OPT and there are two types. One is a glass type core that is not ferrite and the other is the strip core sometimes called blah de blah for marketing reasons. Their B/H curves are almost vertical and very very thin.

In the case of smps, you can only work with a ferrite core because the core must not store energy not even for a microsecond. There really isn't an alternative. To give you idea of what you face, my B+ circuit uses a pair of mosfets in push pull driving a 100 watt core. The push and shove pulses on a 450 volt output is arranged to almost overlap such that snubbers are not required, the reverse energy is recovered and the dv/dt comes out at 900 volts plus to minus in 7 nanoseconds. The clock frequency is nominally 150 kilohertz.

In the case of common mode chokes aka a current compensated choke, there is only ferrite. If you use an iron core for a filament supply, as people do, it will not respond to high frequency noise because the core is too slow. I have seen large copper/iron chokes used here and I believe that users 'hear' an improvement because they will add HF noise into the output.

There are hundreds of ferrite mixes and even if you obtain the tech specs, the vendor does not tell you what the mix actually is so you have to buy them and test yourself.
Ferrite does not work on audio frequencies which is actually a strong benefit.

You mention LNAs, satellite systems. Many years ago, I project managed a 4/6 gig system so I had to learn all that the hard way.
Getting the noise down on a DAC is a challenge. Getting the dual 5 volt supply noise down is also a challenge.
Of some interest, I have been co-operating with a builder of the ESS (Sabre dac) DIY system. The smps drives that rather well. How and why? It is a lot to with the fact that the shunt regulator operates at 500 Kilohertz. The supplied regulator uses a bipolar pass transistor that is just not fast enough.

My filament supplies have changed technology a number of times and I went the way of using commercial converters for the input side because the transformers are very small and are hard on the fingers when winding.

Regards
John

Rerurn to Romy the Cat's Site